CtrlK
BlogDocsLog inGet started
Tessl Logo

giuseppe-trisciuoglio/developer-kit

Comprehensive developer toolkit providing reusable skills for Java/Spring Boot, TypeScript/NestJS/React/Next.js, Python, PHP, AWS CloudFormation, AI/RAG, DevOps, and more.

90

Quality

90%

Does it follow best practices?

Impact

Pending

No eval scenarios have been run

SecuritybySnyk

Risky

Do not use without reviewing

This version of the tile failed moderation
Moderation pipeline encountered an internal error
Overview
Quality
Evals
Security
Files

guide-commands.mdplugins/developer-kit-ai/docs/

AI Plugin — Commands Guide

Overview

The AI plugin provides one command for prompt optimization. It uses advanced techniques (chain-of-thought, few-shot, constitutional AI) to transform basic instructions into production-ready prompts.

Available Commands

CommandDescription
/developer-kit-ai:devkit.prompt-optimizeOptimize prompts at three levels: basic, standard, advanced

/developer-kit-ai:devkit.prompt-optimize

Optimizes prompts using advanced LLM techniques. The command delegates to the prompt-engineering-expert agent.

Syntax

/developer-kit-ai:devkit.prompt-optimize [prompt-text] [target-model] [optimization-level]

Arguments

ArgumentDefaultDescription
prompt-text(required)The prompt text to optimize
target-modelclaude-sonnet-4-6Target LLM model family
optimization-levelstandardDepth of optimization: basic, standard, or advanced

Optimization Levels

LevelTechniques AppliedOutput
basicStructure improvement, clarity enhancements, basic chain-of-thoughtQuick win optimizations
standardCoT, few-shot examples, safety principles, structured outputComprehensive prompt with report
advancedFull optimization suite + testing framework, A/B validation strategyProduction-ready prompt with deployment guide

Output

The command produces three deliverables saved to the working directory:

  1. optimized-prompt.md — Complete prompt text ready for immediate use
  2. Optimization Report — Original assessment, applied techniques, impact metrics
  3. Implementation Guidelines — Model parameters, safety considerations, monitoring recommendations

Examples

# Basic optimization (default model, default level)
/developer-kit-ai:devkit.prompt-optimize "Analyze this code and suggest improvements"

# Standard optimization for Claude Sonnet
/developer-kit-ai:devkit.prompt-optimize "Write a function to process orders" claude-sonnet-4-6 standard

# Advanced optimization for GPT-4 (production-ready)
/developer-kit-ai:devkit.prompt-optimize "Create a code review system" gpt-4 advanced

# Advanced optimization with explicit model
/developer-kit-ai:devkit.prompt-optimize "Build a customer support AI" claude-sonnet-4-6 advanced

Specialized Patterns by Task Type

Task TypeApplied Techniques
Document AnalysisRAG integration, source citation, cross-reference extraction
Code ComprehensionArchitecture patterns, security detection, refactoring recommendations
Multi-Step ReasoningTree-of-thoughts, self-consistency verification, error recovery
ClassificationFew-shot with edge cases, confidence scoring, calibration prompts

Common Workflows

Prompt Iteration Workflow

# 1. Start with basic to get quick wins
/developer-kit-ai:devkit.prompt-optimize "Your initial prompt" claude-sonnet-4-6 basic

# 2. Review the output and identify gaps

# 3. Upgrade to advanced with refined requirements
/developer-kit-ai:devkit.prompt-optimize "Refined prompt with more context" claude-sonnet-4-6 advanced

# 4. Test the optimized prompt with real inputs

# 5. Iterate based on test results

Production Prompt Development

# 1. Define prompt requirements in natural language
# 2. Run advanced optimization
/developer-kit-ai:devkit.prompt-optimize "Production prompt description" claude-sonnet-4-6 advanced

# 3. Review the optimization report
# 4. Test with edge cases and adversarial inputs
# 5. Deploy with monitoring and A/B testing framework

Best Practices

  1. Start with standard — basic is useful for quick tweaks, but standard gives you the full optimization report
  2. Specify the target model — Different models benefit from different techniques; always set this explicitly
  3. Provide context — The more specific your prompt description, the better the optimization
  4. Review all three outputs — The optimized prompt, the report, and the implementation guidelines are all useful
  5. Test with domain inputs — Validate the optimized prompt on representative data before production
  6. Iterate — Use the report's recommendations to refine and re-optimize

Command Selection Guide

TaskCommand
Quick prompt improvement/developer-kit-ai:devkit.prompt-optimize with basic
Comprehensive optimization/developer-kit-ai:devkit.prompt-optimize with standard
Production-ready prompt/developer-kit-ai:devkit.prompt-optimize with advanced
Model-specific tuningSet [target-model] argument explicitly
Adding CoT / few-shotUse standard or advanced level

See Also

  • AI Agents Guide — prompt-engineering-expert agent
  • Prompt Engineering Skill — Skill with reference files for CoT, few-shot, templates, optimization
  • RAG Skill — RAG pipeline implementation for document-grounded prompts
  • Chunking Strategy Skill — Document preprocessing for context-window optimization
  • Core Command Guide — All commands across plugins
  • Java Plugin — LangChain4j Guide — LangChain4j RAG integration

plugins

developer-kit-ai

docs

guide-agents.md

guide-commands.md

README.md

README.md

CHANGELOG.md

context7.json

CONTRIBUTING.md

README_CN.md

README_ES.md

README_IT.md

README.md

tessl.json

tile.json